Baoran wrote on 2020-11-17, 08:19:

True, but I thought it is better to be safe than sorry. Also if someone has cards with diffent ramdacs they can choose the one that has the same ramdac just in case.

BIOS is just software, it's extremely unlikely that mismatched BIOS can do any harm to the hardware.
But it's not a good idea to remove PLCC chips without dedicated PLCC extractor tool - you better get one if you don't have it yet.

Grzyb wrote on 2020-11-17, 10:48:

Baoran wrote on 2020-11-17, 08:19:

True, but I thought it is better to be safe than sorry. Also if someone has cards with diffent ramdacs they can choose the one that has the same ramdac just in case.

BIOS is just software, it's extremely unlikely that mismatched BIOS can do any harm to the hardware.
But it's not a good idea to remove PLCC chips without dedicated PLCC extractor tool - you better get one if you don't have it yet.

Actually one of those extractor tools came with the programmer when I opened the package now. Probably a cheap Chinese one though.

My first attempt reading 5.01 bios chips that are from card with SS2410 ramdac. Is there any way to figure out if I have done this correctly since this is my first time playing with these programmers?

Baoran wrote on 2020-11-17, 12:05:

My first attempt reading 5.01 bios chips that are from card with SS2410 ramdac. Is there any way to figure out if I have done this correctly since this is my first time playing with these programmers?

I dumped my 5.01 BIOS and I get the exact same contents as you Baoran - my card uses the SC11483CV80 DAC

Fascinating...
Now it would be worth to see how does that BIOS behave when it's running with the SC11483 (15-bpp), but some software tries to enter a 24-bpp mode - returns some error, or blindly sets the mode resulting with garbage?
The former would suggest there's some run-time RAMDAC detection, the latter that the BIOS is actually built for the 24-bpp RAMDAC.

Since it seems I have done it correctly. Here is the 6.01 bios version from a card with the same SS2410 chip.

Grzyb wrote on 2020-11-17, 03:53:

KDA0478 is also 8-bpp, but from 24-bit palette […]
Show full quote

imi wrote on 2020-11-17, 03:28:

what is the KDA0478 then?

KDA0478 is also 8-bpp, but from 24-bit palette

Anyway, both KDA0476 and KDA0478 take 8 bits of pixel data at the input, ie. 256 colors
but the pixel data doesn't describe the color directly - there's no R, G, B values there
RGB values are kept in the RAMDAC's palette RAM - 256 x (6-bit R + 6-bit G + 6-bit B) for KDA0476, or 256 x (8-bit R + 8-bit G + 8-bit B) for KDA0478
the pixel data selects which of the palette RAM entry is to be send to the DACs, and the DACs convert the 3 x 6-bit or 3 x 8-bit to analog RGB values

So, there can be 262144 or 16777216 colors, but not simultaneously.

I'm curious what is the deciding factor to look for in the datasheet then just the SRAM size? ^^
also isn't it possible to bypass the SRAM alltogether and drive the DACs directly?

Baoran wrote on 2020-11-17, 14:25:

Since it seems I have done it correctly. Here is the 6.01 bios version from a card with the same SS2410 chip.

this also matches exactly the 6.01 image that imi posted on page 2 of the thread - and he has the samsung DAC

I'd say the "BIOS is DAC specific" myth is now busted

imi wrote on 2020-11-17, 14:51:

I'm curious what is the deciding factor to look for in the datasheet then just the SRAM size? ^^
also isn't it possible to bypass the SRAM alltogether and drive the DACs directly?

With 8-bpp RAMDACs - eg. KDA0476 or KDA0478 - it isn't possible to bypass the palette SRAM.
It's possible with HiColor and TrueColor RAMDACs, eg. SC11483:

maxtherabbit wrote on 2020-11-17, 15:46:

I'd say the "BIOS is DAC specific" myth is now busted

Indeed.
Now I'm wondering about the meaning of the "SpeedSTAR24" name...
I suspect the card was designed for 24-bpp since the very beginning, and even though some cards were shipped with 8-bpp or 15-bpp RAMDACs, there was always the upgrade option.
FCC ID "FTUSPEEDD" registration date is 1992-02-10 - I think 24-bpp RAMDACs were already readily available.

Note that the ET4000 datasheet, from 1990, hardly mentions 16-bpp modes, and doesn't mention 24-bpp modes at all - must have been too early for appropriate RAMDACs...

Grzyb wrote on 2020-11-17, 16:55:

Indeed. Now I'm wondering about the meaning of the "SpeedSTAR24" name... I suspect the card was designed for 24-bpp since the ve […]
Show full quote

maxtherabbit wrote on 2020-11-17, 15:46:

I'd say the "BIOS is DAC specific" myth is now busted

Indeed.
Now I'm wondering about the meaning of the "SpeedSTAR24" name...
I suspect the card was designed for 24-bpp since the very beginning, and even though some cards were shipped with 8-bpp or 15-bpp RAMDACs, there was always the upgrade option.
FCC ID "FTUSPEEDD" registration date is 1992-02-10 - I think 24-bpp RAMDACs were already readily available.

Note that the ET4000 datasheet, from 1990, hardly mentions 16-bpp modes, and doesn't mention 24-bpp modes at all - must have been too early for appropriate RAMDACs...

The utilities/dos drivers that seem to come with ET4000AX cards don't seem to supports higher color vesa modes than 8bpp no matter what ramdac you have and that applies to the version of them that comes with speedstar24 cards. Only way I have been able to get the higher color vesa modes to work in dos seem to be univbe. It would be nice if a version of vmode would exist that would support them that would have come with newer et4000ax card since they were made for quite long time.

I have done one test: have a speedstar VGA with BIOS 4.23 and RAMDAC SC11486CV (32768 colors), removed the SC11486CV RAMDAC and put a
SS2410 RAMDAC (16 million colors), then the 32768 colors does not work, this means BIOS versions 5.01 and 6.01 have code to deal with more
modern RAMDACs like SS2410

This is the last Bios version I have. It is 5.00 from a card with SC11483CV80 ramdac. I had actually some trouble with this because it seems that dbios-H 5.00 chip was made by different manufacturer than the 5 others that I have done. With all the others the software was happy when I set chip type to ST manufacturer and M27C256, but with that chip I had to set it to philips and 27C256 before the software was happy with the device ID it read from the chip.

gfernval wrote on 2020-11-18, 00:21:

I have done one test: have a speedstar VGA with BIOS 4.23 and RAMDAC SC11486CV (32768 colors), removed the SC11486CV RAMDAC and put a
SS2410 RAMDAC (16 million colors), then the 32768 colors does not work, this means BIOS versions 5.01 and 6.01 have code to deal with more
modern RAMDACs like SS2410

Also the cards that I have seen with 4.23 bios seem to have a different layout compared to others and they had also one of the connectors missing too. I don't remember if it was VESA or the multimedia connector that was missing.

Baoran wrote on 2020-11-18, 00:14:

The utilities/dos drivers that seem to come with ET4000AX cards don't seem to supports higher color vesa modes than 8bpp no matter what ramdac you have and that applies to the version of them that comes with speedstar24 cards. Only way I have been able to get the higher color vesa modes to work in dos seem to be univbe. It would be nice if a version of vmode would exist that would support them that would have come with newer et4000ax card since they were made for quite long time.

There's software that can use >8-bpp via non-VBE modes, see eg. QPV/386 1.7e :

What about windows 3.x drivers? The ET4k package I have has options for 32k and 64k colors but neither worked with my AT&T DAC I tried. Of course, this was using the version 4 BIOS so an update may resolve that. Can those of you with the Speedstar24 working in >256 color mode on 3.x chime in with what drivers you're using?

maxtherabbit wrote on 2020-11-18, 04:46:

What about windows 3.x drivers? The ET4k package I have has options for 32k and 64k colors but neither worked with my AT&T DAC I tried. Of course, this was using the version 4 BIOS so an update may resolve that. Can those of you with the Speedstar24 working in >256 color mode on 3.x chime in with what drivers you're using?

I have not tried installing win 3.x on my dos PCs but there is a speedstar24 specific drivers and utilities which includes win 3.x drivers so theoretically those should work.
ftp://retronn.de/driver/Diamond/DIAMOND_SPEED … N_31_DRIVER.exe

That is where I found the speedstar24 specific vmode utility too that allowed me to turn on vesa modes in dos.

It should be noted that various BIOSes provide the 640x480x16M mode via different numbers:

1--------V-1010E0-----------------------------
2INT 10 - VIDEO - Diamond Speedstar 24 - SET 24-BIT GRAPHICS MODE
3        AX = 10E0h
4        BL = video mode (see also #00010 at AH=00h)
5            2Eh = 640x480
6Return: ???
7SeeAlso: AH=00h,AX=10F0h
8--------V-1010F0-----------------------------
9INT 10 - VIDEO - Tseng ET-4000 BIOS - SET HiColor GRAPHICS MODE
10        AX = 10F0h
11        BL = video mode (see also #00010 at AH=00h)
12            13h = 320x200x32K
13            2Dh = 640x350x32K
14            2Eh = 640x480x32K
15            2Fh = 640x400x32K
16            30h = 800x600x32K
17            3Eh = 640x480x16M (Genoa 7900)
18            FFh Tseng 24-bit color mode
19                BH = mode
20                    2Dh = 640x480
21                    2Eh = 640x480x16M
22                    2Fh = 640x400
23                    30h = 800x600
24                    38h = 1024x768??? (Tseng ET4000/W32i)
25Return: AL = 10h if supported
26        AH = status
27            00h if successful
28            other on error
29Desc:   switch the display into a graphics mode with 15 or more bits per pixel
30Note:   the Tseng HiColor BIOS extensions are supported by:
31          Diamond Computer Systems      SpeedStar HiColor VGA
32          Everex Systems                HC VGA
33          Focus Information Systems     2theMax 4000
34          Cardinal Technologies         VGA732
35          Orchid ProDesigner IIs        Genoa 7900
36SeeAlso: AH=00h,AX=10E0h,AX=10F1h,AX=10F2h
37Index:  video modes;Tseng Hi-Color

Baoran wrote on 2020-11-18, 05:01:

That is where I found the speedstar24 specific vmode utility too that allowed me to turn on vesa modes in dos.

Were you able to get any info on what that three way jumper does on the Tseng?

I'm very keen to know if I have a wait state problem, because I had recently tested an 8900c, 8900D, Mach64, and the Tseng 4000. For now only with Doom FPS, on a 486 DLC/40 Mhz - the 8900D is killing it, the C being poor, and the Tseng and Mach 64 pretty much a tie in 2nd place. Weirdest thing is that both 8900s have 80ns, whereas the Tseng and Mach64 have 60ns.

More testing will have to take place.

I have the same 5.01 BIOS rev card as pictured earlier in this thread. What's this three way jumper for?

I have a Speedstar 24X REV A5 with the WDC chip and no BIOS ICs (I have a programmer and EPROMs). Can somebody dump me the BIOSes for this card? I could not find it anywhere yet. Google photo search shows cards with BIOS version 1.04, so it may be a complete different branch than those I found here. I tried one, no success.
Thanks!